<scp>GmFLS2</scp> contributes to soybean resistance to <i>Ralstonia solanacearum</i>

Bacterial strains within the Ralstonia solanacearum species complex (RSSC) are collectively able to cause disease in > 250 plant species from > 50 families (Denny, 2006) and have been recently divided into three species (Safni et al., 2014; Prior et al., 2016): R. solanacearum, R. pseudosolanacearum, and R. syzygii. Most Ralstonia strains are soilborne and penetrate plants through the roots, although some R. syzygii exceptions can be transmitted by insects (Denny, 2006). Upon plant invasion, Ralstonia colonizes plant xylem vessels and multiplies massively, causing a reduction in growth and yield, wilting, and, ultimately, death (Denny, 2006; Xue et al., 2020). The collapse of a diseased plant, which can host > 108 colony-forming units (CFU) per gram of tissue, constitutes a re-inoculation of bacteria into nearby soil, where Ralstonia can survive for years. Ralstonia can then be transmitted by water or other means to other host plants, which can be invaded through natural root openings or directly through wounds caused by other organisms or agricultural practices, such as the use of contaminated tools (Denny, 2006). Strains within the RSSC are the causal agents of devastating diseases in a broad range of economically important crop plants, such as bacterial wilt disease in diverse Solanaceae plants (such as tomato, eggplant, or pepper), brown rot (a.k.a. bacterial wilt) disease in potato, or Moko/blood disease in banana and plantain (Denny, 2006). Due to its persistence, lethality, world-wide distribution, and wide host range, Ralstonia is considered one of the most destructive plant pathogens and a serious threat to food security. The first layer of pathogen perception by plant cells relies on the detection of highly conserved microbial molecules, termed pathogen-associated molecular patterns (PAMPs) by plasma membrane-localized pattern recognition receptors (PRRs; Boutrot & Zipfel, 2017). PRR activation leads to subsequent signaling events and immune responses, ultimately causing PAMP-triggered (or PRR-mediated) immunity (PTI). The biotechnological use of PRRs to engineer plant disease resistance is an emerging approach to fight against plant disease in a wide variety of crop plants and is therefore a promising strategy to contribute to food security world-wide (Lacombe et al., 2010; Mendes et al., 2010; Afroz et al., 2011; Bouwmeester et al., 2014; Tripathi et al., 2014; Albert et al., 2015; Du et al., 2015; Lu et al., 2015; Schoonbeek et al., 2015; Schwessinger et al., 2015; Hao et al., 2016; Kunwar et al., 2018; Omar et al., 2018; Thomas et al., 2018; Mitre et al., 2021). The 22-amino acid epitope flg22, present in bacterial flagellin, is one of the best-studied PAMPs and is perceived by the receptor FLAGELLIN SENSING2 (FLS2; Boutrot & Zipfel, 2017). FLS2 is a leucine-rich repeat-containing receptor-like kinase (LRR-RLK), including an extracellular LRR domain, which mediates ligand-binding, a single-pass transmembrane domain, and an intracellular kinase domain (Sun et al., 2013). The binding of flg22 is mediated by a specific region in the extracellular domain, ranging from the LRR3 to the LRR16 (Sun et al., 2013). FLS2 was originally identified in the model plant Arabidopsis thaliana (hereafter, Arabidopsis), and most angiosperms harbor an FLS2 ortholog that confers flg22 recognition (Felix et al., 1999; Gómez-Gómez & Boller, 2000). Although the ability to perceive bacterial flagellin/flg22 from most bacterial species is conserved in most plant species, there are exceptions. One of those exceptions is the flg22 peptide from Ralstonia, which includes polymorphisms that avoid perception by FLS2 from most plant species, including Arabidopsis and most susceptible crop species, while keeping a functional flagellin (Pfund et al., 2004; Sun et al., 2013; Wei et al., 2018, 2020). The soybean (Glycine max) genome includes two orthologs of Arabidopsis FLS2 (named GmFLS2a and GmFLS2b; Tian et al., 2020). Tomato, a susceptible host for numerous Ralstonia strains, also has two FLS2 paralogs (named SlFLS2.1 and SlFLS2.2), although only SlFLS2.1 encodes a functional flg22 receptor (Jacobs et al., 2017). Conversely, we recently found that both GmFLS2 paralogs encode receptors with an exceptional flg22-binding domain that allows the perception of the polymorphic flg22 from Ralstonia, while keeping its ability to perceive other 'canonical' flg22 peptides from other bacterial pathogens (Wei et al., 2020). The expression of GmFLS2 in leaves of the model Solanaceae Nicotiana benthamiana or roots of susceptible tomato plants leads to enhanced resistance to Ralstonia (Wei et al., 2020), showing the potential of the perception of Ralstonia flg22 by GmFLS2 in mediating disease resistance. Interestingly, despite the fact that Ralstonia strains can infect several legume crops, soybean seems naturally resistant to Ralstonia, since no major incidence of bacterial wilt has been reported in soybean, despite the apparent overlap in their geographical distribution (Denny, 2006; Jiang et al., 2017). In order to determine the contribution of GmFLS2a/b to soybean resistance against Ralstonia, we first generated soybean fls2a/b mutant plants using CRISPR-Cas9-mediated genome editing. Two single-guide RNAs (sgRNAs) were designed to target a conserved region of the GmFLS2a/b genes that encodes the predicted domain responsible for flg22 binding (Sun et al., 2013; Wei et al., 2020; Supporting Information Fig. S1a; Methods S1; Table S1). A single Cas9 vector containing both sgRNAs (Fig. S1b; Mao et al., 2013) was transformed into soybean Williams 82 plants. Among the resulting antibiotic-resistant soybean plants, we isolated two independent lines containing mutations in both GmFLS2a and GmFLS2b genes (hereafter referred to as lines Gmfls2#1 and Gmfls2#2; Fig. S2b). In both independent lines and for both genes, the mutagenesis caused different nucleotide substitutions or deletions leading to early stop codons within the first 800 nucleotides in the GmFLS2 mRNA (Figs 1a, S2). All mutant versions encode predicted short proteins with lengths between 160 and 250 amino acids, containing minimal portions of the flg22-binding region, and lacking most of the LRRs responsible for flg22 binding, the transmembrane domain, and the cytoplasmic domain (Fig. 1b). However, analysis of the predicted mRNAs of the mutant GmFLS2 genes identified additional ATG codons in different open reading frames downstream of the mutated regions (Figs 1b, S2b). If stable, the resulting RNAs would encode truncated FLS2 proteins including a portion of the extracellular domain (but lacking most of the flg22-binding region), the transmembrane domain, and the cytoplasmic domain (Fig. 1b). In order to determine the accumulation of all the mRNAs potentially generated in the mutant lines, we performed quantitative reverse-transcription polymerase chain reaction (qRT-PCR) with different sets of primers (Figs 1c, S2b): primers targeting the region upstream of the mutation in GmFLS2a (GmFLS2a-Up); primers targeting the region upstream of the mutation in GmFLS2b (GmFLS2b-Up); and primers targeting the region downstream of the mutation in both genes, since it was not possible to design specific primers to distinguish between them (GmFLS2-Down). The qRT-PCR results showed a reduction in both transcripts from both genes in leaves of both mutant lines (Fig. 1c). Interestingly, despite the fact that DNA sequencing results were identical in roots and shoots (Figs 1a, S2), all the transcripts were more stable in root tissues, compared with leaves (Fig. 1c), suggesting that the regulation of the stability of these truncated RNAs may be different among plant tissues. In root tissues, in line Gmfls2#1, both transcripts showed reduced accumulation compared with wild-type plants (Fig. 1c); in line Gmfls2#2, however, both transcripts showed similar accumulation to those in wild-type plants (Fig. 1c). Although both transcripts would encode truncated proteins lacking essential domains for their appropriate function, these results should be kept in mind for the interpretation of the performance of both mutant lines. Both mutant lines showed apparently normal growth and development, similar to wild-type plants, when growing under standard laboratory conditions, either in peat pellets sterilized by high-temperature flash drying (Jiffy pots) or commercial potting soil (Fig. S3). To evaluate the loss of function of GmFLS2 in the mutant lines, we monitored the production of reactive oxygen species (ROS) and the activation of mitogen-activated protein kinases (MAPKs) upon treatment with a 'canonical' flg22 peptide (from Pseudomonas syringae, flg22Psy), and the polymorphic flg22 from the R. solanacearum GMI1000 reference strain (flg22Rso) (Methods S1). In leaf tissues, GmFLS2 mutations in both lines caused an abolishment of both ROS production (Fig. 2a–c) and MAPK activation (Fig. S4) upon treatment with either flg22 version, indicating that early flg22-triggered signaling is compromised in both mutant lines. However, in root tissues, although the early ROS burst (peaking c. 15 min after flg22 treatment) was also compromised in both mutant lines, we observed a late ROS burst, peaking c. 60 min after flg22 treatment, in the mutant line Gmfls2#2 (Fig. 2d–f). Upon soil-drenching inoculation with GMI1000, none of the plants, either wild-type or mutant lines, showed any detectable disease symptoms (Fig. S5), suggesting that, under laboratory conditions, GmFLS2 is not required for resistance against Ralstonia present in the soil. However, upon injection of GMI1000 into the stem, wild-type soybean plants showed slight disease symptoms (Fig. 2g). Interestingly, disease symptoms in both mutant lines were stronger than those in wild-type plants, and the mortality of plants was particularly high, and statistically significant, in the mutant line Gmfls2#1 (Fig. 2g,h). Our results indicate that the mutation of GmFLS2a and GmFLS2b enhances disease susceptibility to Ralstonia. It is noteworthy that, although both lines showed an abolishment of early signaling triggered by flg22 in leaf tissues, line Gmfls2#2 displayed a delayed ROS burst after treatment with either flg22 version in root tissues (Fig. 2a–f). Accordingly, line Gmfls2#1 showed a much stronger susceptibility phenotype than line Gmfls2#2 (Fig. 2g,h). In this regard, it is worth remembering that the truncated GmFLS2 transcripts in root tissues of line Gmfls2#2 showed enhanced stability compared with line Gmfls2#1 (Fig. 1b,c), which could potentially result in a stronger accumulation of truncated GmFLS2 proteins in specific cell types or tissues in line Gmfls2#2. In such a situation, we cannot rule out the possibility that the transcript downstream of the mutation generates a truncated GmFLS2 receptor without efficient ligand-binding activity (hence lacking the fast activation of early responses), but that somehow contributes to sustained downstream immune signaling that supports disease resistance. Interestingly, although wild-type plants showed a strong resistance upon both soil-drenching and stem injection with Ralstonia, gmfls2 mutant lines only showed stronger disease symptoms upon stem injection (Fig. 2g,h). This suggests that soybean plants may have additional defense mechanisms that prevent Ralstonia invasion from the soil or its proliferation in root tissues, such as those previously identified in roots of other legume plants (Tran et al., 2016), and the contribution of GmFLS2 may not be significant in such circumstances. However, upon stem injection, both mutant lines showed enhanced susceptibility, suggesting that, once Ralstonia is inside plant tissues, GmFLS2 contributes significantly to soybean defense against Ralstonia, likely by perceiving the polymorphic flg22Rso and initiating defense signaling. Therefore, beyond our observations in the laboratory, GmFLS2 may also contribute to soybean resistance to Ralstonia in agricultural systems, where bacteria may be 'inoculated' into plant tissues by natural means (e.g. caused by other biotic or abiotic factors) or human practices. We thank Rosa Lozano-Duran and Gang Yu for critical reading of this manuscript and helpful discussions, Xinyu Jian and Fangyuan Wu for technical and administrative assistance during this work, Mei Chen for assistance with soybean transformation, and all the members of the Macho and Lozano-Duran laboratories for helpful discussions. This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (grant XDB27040204), the Chinese 1000 Talents Program, and the Shanghai Center for Plant Stress Biology (Center for Excellence in Molecular Plant Sciences, Chinese Academy of Sciences). None declared. APM planned and designed the project. YC, AZ and YW performed experiments. YM and J-KZ designed and carried out soybean mutagenesis. YC and APM analyzed data. APM wrote the manuscript. All the data and materials integral to this study are available within the article and the Supporting Information. Fig. S1 Strategy for the CRISPR/Cas9-mediated mutagenesis of GmFLS2a and GmFLS2b. Fig. S2 Resulting mutations in GmFLS2a/b after CRISPR/Cas9-mediated mutagenesis. Fig. S3 Soybean Gmfls2 mutant lines do not show obvious developmental phenotypes. Fig. S4 Gmfls2 mutant lines show absence of MAPK activation after flg22 treatment. Fig. S5 Soybean Gmfls2 mutant lines do not show disease symptoms upon soil-drenching inoculation. Methods S1 Materials and methods used in this work. Table S1 Primers used in this study. Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.